The Week Orbital Compute Became Real
Readability (Flesch-Kincaid): est. Grade 11 / 10-13 targetAvg sentence length: est. 17 words / 15-22 targetPassive voice ratio: est. 8% / below 15% targetNTP claim scan: 19 e-type sourced / 7 n-type / 0 flaggedLikely-rated items (Findings 6, 7) qualifiedthroughout with "reports indicate" and "analystsnote" languagecriteria clear
Navigation: Space AI Monday #4: The Orbital Cybersecurity Kill Chain | Space AI Monday #6: Coming soon
In fourteen days, between 16 March and 30 March 2026, four independent actors announced developments that, taken individually, each carry strategic weight. Taken together, they constitute something more precise: market confirmation that orbital compute infrastructure has crossed from speculative category to investable asset class.
Starcloud closed a US$170 million Series A at a US$1.1 billion valuation, seventeen months from its Y Combinator demo day. SpaceX filed with the Federal Communications Commission (FCC) for authorisation to operate up to one million orbital data centre satellites. NVIDIA launched its Space Computing platform at GTC 2026, naming six commercial orbital partners. Rocket Lab secured German foreign direct investment (FDI) approval for its Mynaric acquisition, establishing "Rocket Lab Europe" in Munich and positioning for a €6 billion European sovereign constellation contract.
None of these are speculative announcements. They are a commercial investment round, a regulatory filing, a product launch with named partners, and a government-approved acquisition. The market is not discussing orbital compute. It is deploying capital into it.
The Investment Thesis Is Confirmed: Starcloud at US$1.1 Billion
Starcloud was founded in January 2024 and reached unicorn status seventeen months after its Y Combinator demo day: the fastest unicorn trajectory in Y Combinator's history. [1]
The Series A round, closed on 30 March 2026, raised US$170 million at a US$1.1 billion valuation. Benchmark led, with Chetan Puttagunta joining the board. EQT co-led; EQT manages over US$100 billion in assets and operates more than 70 terrestrial data centres across Europe and North America. Additional participants included Macquarie Capital (managing approximately US$500 billion in assets), NFX, Y Combinator, and 776 Ventures. [1] The angel cohort includes a retired US Air Force general, a former Boeing chief executive, and a former Starbucks chief executive.
The investor profile is the analytical signal. EQT does not enter speculative categories. Macquarie manages infrastructure at industrial scale. The angel cohort represents aerospace, defence, and consumer supply chain backgrounds. Together, they have assessed the Corporate Kardashev thesis as investable: the argument that Earth's regulatory and physical constraints, including permitting bottlenecks, local opposition to energy infrastructure, power grid congestion, and cooling water scarcity, are problems to escape rather than solve. Starcloud's founder stated directly that they are "running out of places to build new energy projects on Earth." [1]
This is not a technical claim. It is a governance arbitrage argument: jurisdictions with limited reach over orbital altitude cannot impose the same constraints that terrestrial regulators apply. Benchmark and EQT have now validated that argument with institutional capital.
Starcloud's first satellite, launched in November 2025, carried an NVIDIA H100 into orbit and completed the first orbital AI training run using NanoGPT, along with the first orbital Gemini inference pass. [1] Starcloud-2, scheduled for launch in late 2026, is designed to carry a Blackwell B200 with 100 times the power generation capacity of its predecessor, along with AWS Outposts infrastructure and bitcoin mining application-specific integrated circuits (ASICs). [1] Starcloud-3, currently in development for Starship deployment, is designed to target approximately US$0.05 per kilowatt-hour; according to Starcloud's projections, each Starship launch could carry around 50 Starcloud-3 spacecraft. [1] These are manufacturer projections and are contingent on Starship reaching commercial cadence, which analysts estimate may occur around 2028 to 2029.
The long-term target is tens of gigawatts of orbital compute capacity annually. Whether that specific target is reached matters less than the fact that investors with infrastructure track records have assessed the direction as worth funding at scale.
The Regulatory Battleground: SpaceX and the Million-Satellite Filing
In mid-March 2026, SpaceX filed with the FCC for authorisation to launch and operate up to one million orbital data centre satellites, at altitudes between 500 and 2,000 kilometres, clustered at 50-kilometre intervals to serve varied workload and latency requirements. [2] The satellites would connect via high-bandwidth optical links to the existing Starlink constellation and operate on solar power for over 99% of their operational life.
This filing is a regulatory move inside an active contest. Amazon's Kuiper programme petitioned the FCC to reject the SpaceX application in early March. Blue Origin submitted its own Project Sunrise application for 51,600 satellites in mid-March. [2] FCC Chairman Brendan Carr publicly criticised Amazon's approach.
The Pirate Radio Parallel I traced in Space Mafia is now visible in regulatory proceedings. In the 1960s, radio ships operating from international waters forced national broadcasting authorities into a series of reactive contests. The FCC is now playing the equivalent role: adjudicating between commercial actors each seeking to define the rules before the rules define them. The actor that secures FCC authorisation at orbital scale first will have established a regulatory precedent that shapes data centre governance at altitude for a generation.
For organisations that depend on connectivity, cloud infrastructure, or data processing at any scale, the FCC proceedings are the governance event to watch, not for their immediate operational impact, but for the jurisdictional frameworks they will establish: the questions of data residency, security obligations, and accountability that enterprise leaders will need to answer as orbital compute becomes part of the supply chain.
The combined SpaceX-xAI entity, with a reported combined valuation of approximately US$1.25 trillion, is now seeking to control AI, compute, launch capacity, orbital infrastructure, connectivity, and manufacturing in a single vertically integrated structure. [3] That is the Vertical Integration Singularity: the first entity that controls the full stack, from model to silicon to orbit.
Silicon Goes Orbital: NVIDIA's Space Computing Platform
On 16 March 2026, NVIDIA launched its Space Computing platform at GTC 2026, with six named commercial orbital partners: Aetherflux, Axiom Space, Kepler Communications, Planet Labs, Sophia Space, and Starcloud. [3]
The platform comprises four components. The Space-1 Vera Rubin Module is designed for orbital large language model (LLM) and foundation model processing via solar-powered edge inference. The IGX Thor module supports high-performance AI inference in constrained orbital form factors. The Jetson Orin handles edge-based autonomous space operations and sensing. The RTX PRO 6000 Blackwell supports on-demand ground processing and geospatial analysis. NVIDIA's stated strategy is to reduce what it calls "downlink dependency": instead of transmitting raw sensor data to Earth for processing, platforms process data in orbit and transmit insights. Jensen Huang's summary at GTC: "space computing has arrived." [3]
The practical significance for enterprise strategy is that NVIDIA's Space Computing platform uses the same silicon families already deployed in terrestrial AI infrastructure. The H100 flew on Starcloud-1. The Jetson Orin is the module Kepler Communications deployed across its Tranche 1 constellation (covered in Space AI Monday #4: 40 modules, 10 satellites, now commercially operational). The Blackwell architecture underpins both the terrestrial hyperscale buildout and the orbital layer.
This collapses a critical assumption: that orbital and terrestrial compute are architecturally separate. They share silicon, software frameworks, and developer tooling. Organisations already building AI capabilities on NVIDIA infrastructure are, in principle, a software abstraction layer away from deploying those capabilities at orbital altitude. The data residency, sovereignty, and governance questions that flow from that convergence are not orbital questions. They are enterprise architecture questions with an orbital dimension.
Gen AI Tuesday readers will recognise the parallel: the same dual-use infrastructure pattern I examined there, with the same silicon serving both civilian and military workloads, now extends vertically from terrestrial data centres to low-Earth orbit.
Rocket Lab: From Māhia to Munich
Of the four actors whose announcements defined this fortnight, Rocket Lab's are the most consequential for New Zealand. [4]
On 28 March 2026, Rocket Lab launched its 85th mission, "Daughter of the Stars," from Māhia, deploying Celeste LEO-PNT demonstration satellites in the first dedicated European Space Agency (ESA) launch from New Zealand. [4] Within the same week, the German Federal Ministry for Economic Affairs approved the Mynaric acquisition for approximately US$150 million, establishing "Rocket Lab Europe" in Munich. That acquisition positions Rocket Lab to compete for contracts under the €6 billion IRIS2 European sovereign constellation programme and for German military procurement. [4] Separately, Rocket Lab announced a US$190 million hypersonic test launch contract with the US Department of War for over 20 HASTE missions through 2030: the largest launch contract in the company's history. [4]
The Neutron launch vehicle is targeting its maiden flight in Q4 2026 and is designed to carry 13,000 kilograms to low-Earth orbit (15,000 kilograms in expendable configuration). Māhia's launch complex is expanding to process up to three launch vehicles simultaneously. The Auckland Machine Complex, formerly Precision Components Limited, manufactures components for both Electron and Neutron. [4]
The Venus Life Finder mission, an MIT-developed atmospheric probe targeting a Summer 2026 launch from Māhia, would make New Zealand a participant in the first commercially-launched interplanetary science mission. [4] That context sits alongside the IRIS2 positioning: New Zealand's most commercially significant space company now has a European industrial base, a US defence contract book, and a planetary mission in its launch manifest.
The company is simultaneously a launch provider, a satellite manufacturer, a defence contractor, a hypersonic test platform, and a European industrial entity. The commercial depth represented by those five concurrent roles is what makes Rocket Lab's position structurally different from any other space company in the Southern Hemisphere, or indeed in most of the Northern Hemisphere.
The government's increase in the annual permitted launch count from 100 to 1,000, announced in February 2026 following an ecological risk assessment, provides the regulatory certainty that underpins Rocket Lab's Māhia expansion planning. [5] The count increase and the assessment process are administrative facts; they give Rocket Lab's clients the planning horizon commercial launch contracts require.
The Security Layer: Five Country Council on LEO SATCOM
On 24 March 2026, the Five Country Council, comprising the US National Security Agency (NSA), Australia's Australian Signals Directorate (ASD), New Zealand's National Cyber Security Centre (NCSC), and the Canadian Centre for Cyber Security, released a joint advisory: "Securing Space: Cyber Security for Low Earth Orbit Satellite Communications." [6]
New Zealand is a co-author.
The advisory identifies growing attack surfaces across LEO constellations, covering satellites, ground stations, terminals, end-user devices, and supply chains. [6] Key warnings address radio-frequency links susceptible to jamming, spoofing, and interception; security complications from constant satellite handovers; and supply chain dependencies on commercially-sourced components. The advisory recommends a phased security architecture: implement integrity signing before full encryption; deploy phishing-resistant multi-factor authentication; apply zero-trust principles throughout. [6]
The post-quantum cryptography mandate is specific: national security satellites must be quantum-resistant by 2035, and new acquisitions must incorporate quantum-secure algorithms by 2027. [6]
Reports from the same period describe severe GPS and AIS disruption affecting over 1,100 vessels in the Strait of Hormuz following geopolitical strikes, with vessels reportedly spoofed into false positions and maritime traffic significantly degraded. [7] Intelligence assessments are consistent with known GPS spoofing patterns in that region. The episode illustrates the satellite-dependency vulnerability that the Five Country Council advisory addresses: the same orbital infrastructure on which maritime navigation, aviation, and supply chain timing depend is also a viable target for deliberate interference.
The advisory carries particular weight for NZ organisations because the NCSC co-authored it. This is not a foreign intelligence product being adapted for local relevance. It is a New Zealand government contribution to a joint threat assessment. The attack surfaces it describes, including ground station vulnerabilities and supply chain risks, are directly applicable to NZ organisations that operate or depend on satellite communications infrastructure. The EA Thursday series on Zero Trust architecture addresses exactly the principles the advisory recommends at the network layer; the two threads converge on the same defensive architecture for organisations at either altitude.
Two Space Clouds: The Geopolitical Architecture
The four commercial developments above are unfolding inside a geopolitical structure. US commercial orbital compute, led by SpaceX's Starlink constellation combined with Starcloud, Planet Labs, and ICEYE, represents one architectural bloc. China's state-directed Three-Body Computing Constellation, targeting 2,800 satellites and approximately one exaflop of distributed processing capacity, represents a second. [8] The European Union's IRIS2 multi-orbit sovereign constellation, expecting initial operational capability around 2030, represents a third. Rocket Lab Europe's positioning, post-Mynaric, places New Zealand-rooted capability inside that third bloc.
Reports from analysts and PLA-affiliated commentary indicate that China's authorities assess Starlink's expansion trajectory as a national security concern, and that China is responding with two parallel mega-constellation programmes targeting a combined total in the hundreds of thousands of satellites. [8] China launched an initial 12-satellite cluster in May 2025. [8] CASC announced in late March 2026 its intention to develop "gigawatt-level space digital intelligence infrastructure" integrating satellite constellations with AI-powered processing and autonomous resource extraction, targeting a leading space power position by 2045. [8]
Artemis II, with crew members Reid Wiseman, Victor Glover, Christina Koch, and Jeremy Hansen, is in final preparation for its 10-day crewed lunar flyby: the first crewed deep space mission since 1972. [8] The Artemis governance framework and the International Lunar Research Station (ILRS) programme led by China and Russia represent competing architectural standards for the next phase of human activity beyond low-Earth orbit, running in parallel to the commercial orbital infrastructure contest.
The US NASA Authorization Act 2026 mandates two independent commercial stations before the International Space Station's planned retirement in 2030, specifically to prevent what the Act terms "Gap Risk." [8] The provision reflects a structural shift: sovereign space capability is increasingly delivered through commercial procurement rather than direct state operation. That model creates the dependencies and the governance gaps that the Five Fault Lines framework identifies.
What This Means for NZ Organisations
Four specific considerations for NZ enterprise and board leaders.
GBSI Act compliance deadline: 119 days. The Ground-Based Space Infrastructure Act requires ground-based space infrastructure operators to hold formal ministerial authorisation by 29 July 2026. MBIE and NZSIS held a joint webinar on 4 March 2026 covering protective security requirements and due diligence processes. [5] NZSIS has confirmed that NZ ground-based space infrastructure operators have received requests to develop infrastructure for entities seeking to conceal foreign military affiliations; this is the security context the Act was designed to address. [5] If your organisation operates ground station infrastructure of any kind, the compliance clock is running.
GPS dependency is a board-level risk. The Strait of Hormuz spoofing incident is a relevant reference for any NZ organisation whose operations depend on GPS or Automatic Identification System (AIS) positioning. [7] Maritime logistics, aviation, precision agriculture, and supply chain timing systems that depend on GNSS integrity face equivalent disruption risk from deliberate spoofing or jamming. The Five Country Council advisory describes mitigations applicable to ground segment operators. The question for NZ boards is whether GPS dependency appears in their risk registers at the appropriate tier.
Orbital compute economics are a watch item for infrastructure investment. The proposed Datagrid AI factory at Invercargill, with an estimated NZ$3.5 billion investment, represents the class of hyperscale terrestrial facility that the orbital compute thesis would, over a longer horizon, tend to disrupt. The timeline is not confirmed. Whether Starcloud's projected cost-per-kilowatt-hour is achievable depends on Starship commercial cadence and sustained investor confidence through multiple capital cycles. What is confirmed is that investors with infrastructure track records are funding that disruption thesis. NZ infrastructure investment strategy should hold it as a watch item, not a settled outcome.
The silicon stack is converging. NVIDIA's Space Computing platform uses the same silicon families that NZ organisations are already deploying on-premises and via cloud providers. The architectural implications, including questions of data residency, sovereignty obligations under the GBSI Act and the Privacy Act 2020, and the extension of AI governance frameworks to orbital infrastructure, will eventually reach NZ enterprise leaders. The Five Fault Lines I mapped in Space Mafia identify where those questions land: data sovereignty, intellectual property jurisdiction, environmental accountability, consolidation risk, and security. None of those fault lines have been resolved by the announcements of the past fortnight. All of them have been brought materially closer.
The Fortnight in Summary
The Corporate Kardashev thesis has institutional investors. SpaceX is seeking FCC authorisation for a million orbital data centres. NVIDIA's silicon is commercially operational in orbit and named six partners to prove it. New Zealand's most important space company now has a European industrial base, a US defence contract, and a planetary mission in its schedule.
The Five Country Council, with New Zealand as a co-author, has assessed the orbital attack surface and issued specific security architecture guidance. The governance frameworks, the Five Fault Lines identified in Space Mafia, are not closing the gap with commercial deployments. The gap is widening.
Orbital compute is not a future concern for enterprise strategy. The capital has arrived, the hardware is in orbit, and the regulatory contests are underway. The relevant question for organisational leaders is whether their governance frameworks are keeping pace with the investment cycle.
What question about orbital compute governance would you most want answered before your next board or executive meeting?
The views expressed in this article are entirely my own, informed by more than 30 years of professional experience in architecture, security, and technology leadership in New Zealand. They do not represent the views of my employer, any government agency, or the New Zealand government. My commentary on legislation and policy is analytical, drawing on publicly available sources and my professional expertise in architecture, security, and AI governance. I follow the Public Service Commissioner's Code of Conduct for the Public Sector and social media guidance.
Andreas Hamberger is a New Zealand-based technology leader and board director with more than 30 years of experience in enterprise architecture, cybersecurity, and AI governance. He writes The Hamberger Report, a LinkedIn series covering the intersection of technology, governance, and organisational strategy. Space Mafia examines the sovereignty implications of orbital compute infrastructure.
I use AI tools, including Sudowrite, Claude, Perplexity AI, DeepSeek AI, ChatGPT, Grok, Copilot, Openart and Gemini, as deliberate production tools, not ghostwriters. This is consistent with my position: AI amplifies human judgement; it does not replace it. The frameworks, arguments, and editorial decisions in this series are original work. AI accelerated the process. The thinking is mine.
[1] Starcloud / BusinessWire / TechCrunch / SpaceNews / GeekWire / Via Satellite / DataCenterDynamics. "Starcloud Raises US$170M Series A, Achieves Unicorn Status." 30 March 2026. [URL to be inserted before publication]
[2] DataCenterDynamics / SpaceNews. "SpaceX Files FCC Application for Up to One Million Orbital Data Centre Satellites." Mid-March 2026. [URL to be inserted before publication]
[3] NVIDIA Newsroom. "NVIDIA Launches Space Computing Platform at GTC 2026." 16 March 2026. https://nvidianews.nvidia.com/
[4] Rocket Lab / Globe Newswire / SpaceNews / SatNews. "Rocket Lab 85th Mission; Mynaric FDI Approval Confirmed; US$190M Department of War HASTE Contract." March 2026. [URLs to be inserted before publication]
[5] MBIE / NZSIS. "Ground-Based Space Infrastructure Act Compliance Guidance; Launch Limit Increase to 1,000 Annual Launches." New Zealand Government. February-March 2026. [URL to be inserted before publication]
[6] NSA / ASD ACSC / NCSC-NZ / CCCS. "Securing Space: Cyber Security for Low Earth Orbit Satellite Communications." Five Country Council Joint Advisory. 24 March 2026. [URL to be inserted before publication]
[7] Intelligence briefing. "Strait of Hormuz GPS and AIS Disruption Affecting 1,100+ Vessels." March 2026. Confidence: Likely; consistent with established GPS spoofing patterns in the region. [URL to be inserted if primary source confirmed before publication]
[8] SatNews / CASC announcement / CSIS China Power Project / NASA Authorization Act 2026. "Three-Body Computing Constellation; CASC Gigawatt Announcement; Artemis II; NASA Authorization Act 2026." March 2026. [URLs to be inserted before publication; China constellation specifics rated Likely; Artemis II crew and NASA Act confirmed]